trustlog-guard

What This Skill Does

TrustLog Guard is a comprehensive financial governance and observability tool designed specifically for OpenClaw agents. It operates by auditing your local session logs to provide granular insights into API consumption and expenditure. By parsing the .jsonl session files generated by your agent, the skill calculates exact costs per message, tracks trends across models, and identifies high-spending patterns. It empowers users to transition from blind usage to informed resource management, ensuring that every token spent aligns with project objectives.

Installation

To integrate TrustLog Guard into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/anouartrust/trustlog-guard

Use Cases

TrustLog Guard is ideal for developers and AI power users who want to prevent budget overruns while experimenting with high-end models.

• Cost Auditing: Perfect for agencies or freelancers who need to bill clients accurately for AI-assisted research and development.
• Budget Enforcement: Use it to set strict daily or monthly limits, receiving proactive warnings before you hit your financial caps.
• Optimization Strategy: Analyze which specific models, such as Claude-Opus, are depleting your budget and shift simple tasks to more economical alternatives like Claude-Haiku.
• Anomaly Detection: Quickly identify runaway loops or inefficient prompt structures that result in unusually high costs per session.

Example Prompts

1. "/spend - Can you give me a breakdown of my current spend and suggest ways to optimize my model usage?"
2. "/budget set monthly $150 - Please update my monthly budget cap for this agent."
3. "/budget - Check my current status to see how close I am to my daily limit."

Tips & Limitations

• Privacy: TrustLog Guard operates 100% locally. Your session logs are never transmitted to external servers or the skill author.
• Data Integrity: Ensure that your OpenClaw agents are configured to write logs to the standard directory; otherwise, the skill will prompt you to verify the path.
• Optimization: Focus on the 'Top sessions' report to identify heavy prompt structures. Refactoring these can often save more than switching models entirely.
• Scope: It only tracks costs reported within the session logs. If you use external API keys outside of the OpenClaw agent scope, those costs will not be reflected here.